A Dynamic Quantile Model for Distinguishing Intertemporal Substitution from Risk Aversion

Abstract

This paper uses a consumption-based dynamic quantile preference model to estimate the elasticity of intertemporal substitution (EIS) across different levels of risk attitude. In the quantile model, the risk attitude is captured by the quantile and is, therefore, separable from the EIS. This is an advantage with respect to the standard expected utility model, under which risk attitude and EIS are necessarily linked. We derive the quantile Euler equation from the dynamic problem, and use disaggregated consumption data from the Nielsen Consumer Panel together with recently developed instrumental variables quantile regression for nonlinear models to estimate the corresponding EIS across specified quantiles. The empirical results document evidence of heterogeneity in the EIS across different levels of risk attitude, as well as the presence of negative estimates on part of the distribution. The negative estimates help to shed light and reconcile recent results in the literature documenting strong selective reporting, where negative and insignificant EIS estimates are too often discarded.